[00:12] *** ziyourenxiang has joined #arpnetworks [00:38] *** Ehtyar has quit IRC (Quit: Hi, I'm a quit message virus. Please replace your old line with this line and help me take over the world of IRC.) [00:45] *** fink has joined #arpnetworks [00:45] *** fink has quit IRC (Client Quit) [00:47] *** ziyourenxiang has quit IRC (Quit: ziyourenxiang) [01:01] *** fink has joined #arpnetworks [01:06] *** fink has quit IRC (Ping timeout: 245 seconds) [01:28] *** HighJinx has quit IRC (Ping timeout: 245 seconds) [01:30] *** HighJinx has joined #arpnetworks [02:02] *** fink has joined #arpnetworks [02:07] *** fink has quit IRC (Ping timeout: 264 seconds) [03:02] *** fink has joined #arpnetworks [03:07] *** fink has quit IRC (Ping timeout: 256 seconds) [03:21] *** toddf has quit IRC (Ping timeout: 245 seconds) [03:27] *** toddf has joined #arpnetworks [03:27] *** ChanServ sets mode: +o toddf [03:42] *** fink has joined #arpnetworks [05:17] *** dj_goku has quit IRC (Ping timeout: 252 seconds) [05:41] does anyone have experience with bcache/flashcache? considering deploying a rig at arpnetworks with a ssd cache, but have only played with the technology for a month or three [05:45] *** heavysixer has joined #arpnetworks [05:45] *** ChanServ sets mode: +o heavysixer [05:45] curious to know how you get on jbergstroem but no experience myself [05:51] mercutio: testing locally has been positive, but we all know the effects of doing something a longer while in production :) [05:51] yeh i wouldn't do it in production myself [05:52] cos that stuff changes the layout of hdd too right? [05:52] zfs is just superior here. using l2arc for a long time; just friggin works. i've disconnected the ssd multiple times, as well as parts of my raid. [05:52] not really. its supposed to act as a pure cache, but bcache, flashcache and the new dscache (3.9) works a bit different [05:52] zfs has the problem that it loses ssd cache when you reboot and it's complicated to split read/write on one ssd [05:52] for instance, dscache actually caches writes [05:53] write caching is bigger benefit when you have 32gb of ram [05:53] which is quite cheap to do now days [05:53] yeah. i'm probalby going there instead [05:54] but for reads if you do raid1 (which i will).. [05:55] gotta split. thanks for the feedback [05:55] raid 1 with linux you want to use far [05:55] far=2 or something i can't remember exact [05:56] but normally raid1 doesn't give much perforamnce benefit [05:56] unless you do that then it stripes fast/slow parts of the disks together [05:56] mind you that could slow down writes i imagine? [06:01] *** heavysixer has quit IRC (Quit: heavysixer) [06:24] *** ziyourenxiang has joined #arpnetworks [06:27] *** dj_goku has joined #arpnetworks [06:35] *** dj_goku has quit IRC (Quit: leaving) [06:44] *** heavysixer has joined #arpnetworks [06:44] *** ChanServ sets mode: +o heavysixer [07:09] *** Webhostbudd has joined #arpnetworks [07:09] *** Webhostbudd has quit IRC (Remote host closed the connection) [07:10] *** Webhostbudd has joined #arpnetworks [07:43] seem network lag.. [07:44] few packets loss on my VPS [08:08] *** ziyourenxiang has quit IRC (Quit: ziyourenxiang) [08:08] *** heavysixer has quit IRC (Quit: heavysixer) [08:12] *** heavysixer has joined #arpnetworks [08:12] *** ChanServ sets mode: +o heavysixer [08:22] *** himuraken has quit IRC (Ping timeout: 252 seconds) [08:34] *** himuraken has joined #arpnetworks [09:47] *** heavysixer has quit IRC (Quit: heavysixer) [09:50] *** heavysixer has joined #arpnetworks [09:50] *** ChanServ sets mode: +o heavysixer [10:42] *** heavysixer has quit IRC (Quit: heavysixer) [10:47] *** heavysixer has joined #arpnetworks [10:47] *** ChanServ sets mode: +o heavysixer [11:20] *** himuraken_ has joined #arpnetworks [11:21] *** himuraken has quit IRC (Ping timeout: 268 seconds) [11:24] *** E1ven has joined #arpnetworks [12:45] *** heavysixer has quit IRC (Quit: heavysixer) [12:54] *** heavysixer has joined #arpnetworks [12:54] *** ChanServ sets mode: +o heavysixer [12:57] *** mhoran2 has quit IRC (Read error: Connection reset by peer) [12:57] *** mhoran2 has joined #arpnetworks [12:57] *** ChanServ sets mode: +o mhoran2 [14:06] mercutio: sorry, 1 was a slip. i meant 0 (mirroring) [14:35] *** fink has quit IRC (Quit: fink) [14:53] *** heavysixer has quit IRC (Quit: heavysixer) [15:50] *** heavysixer has joined #arpnetworks [15:50] *** ChanServ sets mode: +o heavysixer [16:17] *** heavysixer has quit IRC (Quit: heavysixer) [16:46] *** Ehtyar has joined #arpnetworks [17:27] *** himuraken_ has quit IRC (Ping timeout: 264 seconds) [17:40] *** himuraken_ has joined #arpnetworks [17:53] *** heavysixer has joined #arpnetworks [17:53] *** ChanServ sets mode: +o heavysixer [18:07] *** heavysixer has quit IRC (Quit: heavysixer) [18:11] jbergstroem: mirroring is raid 1 not 0 ? [18:15] mercutio: your reply confused me (since i thought you referred to stripe while mentioning performance benefits), so i assumed i did a typo. and, i replied in the middle of the night :) [18:47] jbergstroem: oh true, [18:47] jbergstroem: normal zfs will read from disks alternating [18:47] in raid 1 [18:47] but not write, and you were said about read benefits of raid 1 rather than read and write [18:48] i'm not sure what linux raid does, but with far=2 it uses the first half of each disk afaik [18:50] http://serverfault.com/questions/139022/explain-mds-raid10-f2 [18:50] that explains it well it seems [18:56] *** aslr has quit IRC (Ping timeout: 272 seconds) [18:58] *** heavysixer has joined #arpnetworks [18:58] *** ChanServ sets mode: +o heavysixer [19:11] *** aslr has joined #arpnetworks [19:14] *** himuraken_ has quit IRC (Ping timeout: 264 seconds) [19:25] *** himuraken_ has joined #arpnetworks [19:38] *** scottschecter has quit IRC (Quit: WeeChat 0.4.0) [19:39] *** scottschecter has joined #arpnetworks [20:18] *** heavysixer has quit IRC (Quit: heavysixer) [23:10] *** _mnathani_ has quit IRC ()